flashlight2537

[quaint fable]

Please be patient

Our volunteers look into many questions every day; sometimes it takes them a little while to answer.

Pro tip: you can rename the thread title with `.tr <thread name>`

Make it descriptive, including relevant context, but also to the point. This way you improve your chances of getting a more relevant and specific answer.

[tulip basalt]

well here's for trusting chatGPT

"Je ne suis pas vivant, mais je grandis, je n’ai pas de poumon, mais j’ai besoin d’air et je meurs sous l’eau" is the actual riddle

don't forget generative AI is not a reliable source

[plucky loom]

Aww man. I've been chatting with this AI for days thinking it was really helping me improve. Now I'm learning it doesn't know what it's talking about...🥲

[tulip basalt]

;chatgpt

[quaint fable]

ChatGPT et les autres grands modèles de langage d'IA sont des outils intéressants pour l'apprentissage des langues, mais faillibles. Les informations qu'ils donnent peuvent être erronées, tout en donnant des justifications plausibles pour quelqu'un ne maîtrisant pas la langue. Il ne faut pas les utiliser comme source ultime d'information. Ils peuvent cependant aider à déconstruire une phrase en divers concepts (que vous pouvez ensuite vérifier), donner des traductions contextualisées, et simuler divers interlocuteurs.

ChatGPT and other large AI language models are interesting tools for language learning, but they are fallible. The information they provide can be incorrect while offering seemingly logical justifications to someone who does not fully master the language. They should not be used as an endpoint for information. However, they can help deconstruct a sentence into various concepts that you can then verify, provide contextualized translations, and simulate diverse conversational partners.